ARTICLE

pubs.acs.org/jced

Vapor-Liquid Equilibria in the Propyl Acetate Ethanoic Acid Binary

System from (323.15 to 353.15) K: Measurement with a Static Method
and Modeling with the NRTL, Wilson, UNIQUAC, and COSMO-SAC
Approaches
Dongfu Fan, Jinlong Li,, Jibin Shi, Changjun Peng,*, Honglai Liu, Ying Hu, and Patrice Paricaud

State Key Laboratory of Chemical Engineering and Department of Chemistry, East China University of Science and Technology,
Shanghai 200237, China

Laboratoire de Chimie et Proc
ed
es, ENSTA-ParisTech, 32 Boulevard Victor, 75739 Paris Cedex 15, France
ABSTRACT: A static total pressure method was employed to determine the vapor pressures of pure propyl ethanoate (propyl
acetate), pure ethanoic acid, and the corresponding propyl acetate ethanoic acid binary system at dierent feeding
compositions over the temperature range (323.15 to 353.15) K. The isothermal vapor-liquid equilibrium (VLE) data of the
binary mixture at various temperatures were obtained from classical thermodynamic relations and mass-balance equations. The
nonrandom two-liquid (NRTL), Wilson, and universal quasi-chemical (UNIQUAC) models were used to represent the
nonideality of the liquid phase, and a modied Peng-Robinson equation of state was used to compute the properties of the vapor
phase. The overall average relative deviations between the experimental equilibrium pressures and the NRTL, Wilson, and
UNIQUAC models were (0.95, 0.96, and 0.94) %, respectively. The three models gave about the same equilibrium pressures and
vapor compositions. A new version of the conductor-like screening model-segment activity coecient (COSMO-SAC) model
was applied to predict the VLE of propyl acetate ethanoic acid mixtures, and good agreement with the experimental data was
obtained.

INTRODUCTION
Ethanoic acid is often used as a raw material for the production
of ethenyl acetate, acetic anhydride, acetate ber, acetic ester, and
so on.1 Mixtures containing ethanoic acid in industrial processes
are usually very complex and consist of many components and
multiple phases, in which some components can form azeotropes
with ethanoic acid. Azeotropic distillation with the help of an
entrainer such as propyl ethanoate (propyl acetate) is always
adopted in separations of such mixtures.2 Some phase equilibrium data of propyl acetate ethanoic acid binary mixtures have
been reported in the literature. For instance, Othmer3 provided
the equilibrium compositions of the vapor and liquid phases of
the propyl acetate ethanoic acid mixture at atmospheric
pressure but did not give the corresponding equilibrium temperature. Fu et al.4 reported isobaric vapor-liquid equilibrium
(VLE) data for the propyl acetate ethanoic acid system at 760
mmHg. Recently, Xiao et al.5 measured the liquid-liquid
equilibrium (LLE) data of the ternary system water ethanoic
acid propyl acetate at T = (298.15, 313.15, and 363.65) K
under atmospheric pressure. Toikka and Toikka6 briey reviewed the VLE and LLE data in quaternary mixtures of ethanoic
acid, propyl acetate, water, and n-propanol and discussed the
azeotropic properties and the topological structure of the residue
curve map at 313.15 K for this system. However, to our knowledge, very few binary VLE data for ethanoic acid and propyl
acetate at low pressures have been reported in the literature. For
that reason, new VLE measurements for binary mixtures of
propyl acetate and ethanoic acid as well as for pure propyl
r 2011 American Chemical Society

acetate and pure ethanoic acid over the T range (323.15 to

353.15) K were performed in the current work using a static total
pressure method. As is well-known, this method has the merit of
avoiding the analysis of equilibrium compositions.
Besides the experimental measurement of phase equilibrium,
the thermodynamic model is also an eective tool for understanding phase behavior, and it is often crucial for the simulation
and optimization of industrial separation processes. Excess Gibbs
models are always applied for the correlation of experimental
data.4,5 In the present work, the nonrandom two-liquid (NRTL),7
Wilson,8 and universal quasi-chemical (UNIQUAC)9 models were
applied to represent the nonideality of the liquid phase. The
vapor phase at low pressure is often assumed to be an ideal
mixture for the compounds without hydrogen bonding, dipoledipole interactions, and so on. However, mixtures containing
ethanoic acids cannot be simply considered as ideal mixtures
because ethanoic acids are associating molecules with strong
hydrogen-bonding interactions. Here, the modied PengRobinson equation of state (MPR EoS) presented by Liu
et al.,10,11 which is based on a large set of experimental data for
associating uids, was used to describe the nonideality of the
vapor phase. In addition, the predictive conductor-like screening
Special Issue: John M. Prausnitz Festschrift
Received: October 29, 2010
Accepted: January 10, 2011
Published: February 02, 2011
1323

dx.doi.org/10.1021/je101131x | J. Chem. Eng. Data 2011, 56, 13231329

Journal of Chemical & Engineering Data

model-segment activity coecient (COSMO-SAC) model
was used to determine the VLE of propyl acetate ethanoic
acid, and the predictive capacity for this system was checked and
conrmed.

ARTICLE

following equations:
R T
a
P 3 V - b V 3 V b b 3 V - b

EXPERIMENTAL SECTION
Chemicals. Propyl acetate (analysis purity grade, minimum
mass fraction 0.99) was purchased from Shanghai Jingchun
Reagent Company and ethanoic acid (analysis purity grade,
minimum mass fraction 0.995) from Shanghai Chemical Regent
Company. Before it was used in experiments, the ethanoic acid
was distilled and recrystallized, and both the propyl acetate and
ethanoic acid samples were dried using 0.4 nm molecular sieves.
The exact purities of propyl acetate and ethanoic acid were
determined by analysis using gas chromatography (Agilent 6890)
with flame ionization detection, and they were found to be not
less than 0.982 and 0.999 in mole fraction, respectively.
Apparatus and Operation Procedure. The experimental
apparatus used in this work was composed mainly of a vacuum
system, a thermostatic system, and a measurement system, and it
has been described elsewhere.12,13 First, a liquid sample with
known total amount nT and total mole fraction zi of component i
was injected into a glass equilibrium still with a volume of 45 cm3.
Next, to remove gas impurities from the liquid sample, the
processes of freezing, degassing, and thawing were carried out
and repeated (usually three times) until the vapor pressure above
the frozen liquid sample was less than about 7 Pa. The equilibrium still was then placed in a water bath at a given temperature,
which was controlled via a platinum resistance thermometer ((
0.05 K). The VLE was considered as reached when no temperature or pressure changes were observed after a given time (usually
1 h), and the equilibrium temperature T, pressure Pexptl, vaporphase volume VG, total amount nT, and total mole fraction zi
were recorded as the original experimental data at equilibrium.
The next experimental measurement was then carried by adjusting the temperature of the water bath to another value. It should
be noted that the equilibrium temperature was measured using a
mercury thermometer with an uncertainty of ( 0.02 K (which
was calibrated by Shanghai Institute of Measurement and Testing
Technology before use) rather than a platinum resistance
thermometer, which was used only to control the temperature
of the water bath. The equilibrium pressure was measured using a
mercury pressure gauge and a cathetometer with an uncertainty
of ( 0.04 mm. Five types of experimental data were determined
experimentally: the total amount nT, the total mole fraction zi of
component i, the equilibrium temperature T and pressure Pexptl,
and the vapor-phase volume VG. The other properties were
determined using a thermodynamic model.

MODELS
The expressions for the NRTL, Wilson, and UNIQUAC
activity coecient models can be found elsewhere,7-9 and are
not repeated here. We provide only the main working equations
for the MPR EoS10,11 and the modied COSMO-SAC model.14
Modified PR Equation of State. The cubic Peng-Robinson
(PR) equation of state15 has a simple form and is often used for
modeling of the phase equilibria of pure fluids or mixtures. To
improve its precision and expand its application, Liu et al.10,11
presented a modified PR EoS (MPR EoS) based on a new
expression for the parameter R(Tr); the MPR EoS is given by the

0:45724 3 R 2 3 Tc 2
3 RTr
Pc

0:07780 3 R 3 Tc
Pc



n
RTr 1 1-Tr 3 m 2
Tr
b

2a
2b
3

where R is the gas constant, Pc and Tc are the critical pressure and
temperature, respectively, and Tr = T/Tc is the reduced temperature. The model parameters m and n are pure-component
parameters that can be determined by fitting the vapor pressures
or the saturated liquid densities of the pure compounds. The
mixing rules are given by the following equations:
XX
xi 3 xj 3 aij
4a
a
i1 j1

X
xi 3 bi

4b

i1

p
aij ai 3 aj

where xi is the liquid phase mole fraction. Finally, ji, the fugacity
coefficient of molecule i in the mixture, is calculated using the
expression:
0 P
1
2 3 xj 3 Aij
Bi
A B
Bi C
j
- A
ln ji 3 Z - 1 - lnZ - B - p 3@
A
B
B
2 23B
"
p #
Z 1 2 3 B
6
3 ln Z 1 - p2 B
3
where A = (a 3 P)/(R 3 T)2, B = (b 3 P)/(R 3 T), Aij = (aij 3 P)/(R 3 T)2,
and Bi = (bi 3 P)/(R 3 T); Z = (P 3 V)/(R 3 T) is the compressibility
factor, which is calculated using:
Z3 B - 1 3 Z2 A - 2 3 B - 3 3 B2 3 Z B2 B3 - A 3 B 0 7

Equation 7 is solved to obtain Z, from which the vapor molar

volume and the fugacity coefficient in the vapor phase can be
computed at a given T and P.
COSMO-SAC Model. COSMO (conductor-like screening
model)-based thermodynamic models were first proposed by
Klamt et al.16-20 The COSMO for real solvents (COSMO-RS)
model developed by Klamt and co-workers16-20 is a predictive
activity coefficient model in which the only needed inputs that
characterize a given compound are ab initio data. The model
depends on universal parameters that were determined over a
large set of experimental phase equilibrium data. Following the
works of Klamt et al.,16-20 Lin and Sandler21 proposed the
COSMO-SAC model based on the COSMO-RS model.16-20
The original COSMO-SAC model has been used extensively to
predict VLE and LLE data.21 Recently, Hsieh et al.14 proposed a
1324

dx.doi.org/10.1021/je101131x |J. Chem. Eng. Data 2011, 56, 13231329

Journal of Chemical & Engineering Data

ARTICLE

from the prole of the uid as

2
nHB
, HB X
X
psj sn 3 sj sn
lntj tm  -ln4

modified version of the COSMO-SAC model in which they

introduced a temperature-dependent parameter for the electrostatic interaction and considered the profiles of different
functional groups that are involved in hydrogen bonding. This
new version of the COSMO-SAC model has more universal
parameters that were fitted to LLE data.14 In this model, the
activity coefficient i/S of a solute i in the solution S is expressed
as
nHB
, HB X
X


n
pi sm 3 lnsS sm 
lni=S lnSG
i3
i=S
s
m

s s
-lni m 
8

-Wtm ,sn
3 exp
RT


10

where the subscript j can refer to either a pure liquid i or the

mixture S, the superscripts t and s represent either a hydrogenbonded (HB)
(nHB) segment,23 and
P or non-hydrogen-bonded
P
molec
) and pi() = Ai()/Amolec
pS() = i[xi 3 Ai()]/ i(xi 3 Ai
i
are the normalized distributions of surface charge densities for
the mixture S and molecule i, respectively, in which Ai() =
nHB
() is the sum of the surface areas of segments
AHB
i () Ai
having a charge density , AHB
i () is the total surface area of the
segments of charge that are around oxygen, nitrogen and
uorine atoms and hydrogen atoms connected to a N, O or F
() is the sum of the areas of all of the other
atom, and AnHB
i
segments of charge . The prole pi() of component i is given
by
nHB
iOT
pi pnHB
pHB
iOH
i
i pi
p p 11

where the ln(SG

i/S) term is the Guggenheim-Staverman combinational term,22 given by




i
z
i
X
SG
lni=S ln
xj 3 lj 9
3 qi 3 ln
li - i 3
2
xi
i
xi j
P
in which i = (xi 3 ri)/ j(xj 3 rj) is the volume fraction
P of solute i,
z = 10 is the coordination number, i = (xi 3 qi)/ j(xj 3 qj) is the
/vref
area fraction, li = z 3 [(ri - qi)/2] - (ri - 1), and ri = vmolec
i
and qi = Amolec
/Aref are the normalized volume and surface
i
and Amolec
are the volume and
area, respectively, where vmolec
i
i
surface area of solute i, respectively, as determined from ab
initio calculations, vref = 6.669 3 10-2 nm3, and Aref = 0.7953 nm2.
/ae is the number of surface segments and
In eq 8, ni = Amolec
i
ae = 0.0725 nm2 is an eective surface area.23 The activity coecient of segment m having charge density m is determined

() is the prole for surface segments that do not

where pnHB
i
contribute to hydrogen bonding and pOH
i () (for OH in water
and alcohol) and pOT
i () (for O in ketones, NO2 in nitro
compounds, and NH2 in amines) are the proles of segments
that contribute to hydrogen bonding. The exchange energy
between surface segments m and n with charge densities m
and n, W(tm, sn), is given by
W tm ,sn cES 3 tm sn 2 - chb tm ,sn 3 tm - sn 2 12
where cES = AES BES/T2 is a temperature-dependent parameter
and chb(tm, sn) is the hydrogen-bonding interaction energy,
which is dened as
8
cOH-OH , s t OH, tm 3 sn <0
>
>
>
<c
s t OT, tm 3 sn <0
OT-OT ,
chb tm , sn
cOH-OT , s OH; t OT, tm 3 sn <0
>
>
>
:0
otherwise
13
The quantities AES, BES, cOH-OH, cOT-OT, and cOH-OT are
adjustable parameters, and their values can be found in the
literature.14 Here only the main working equations have been
given; for further detailed information, see the papers by Lin and
co-workers.14,21-25

Figure 1. Comparison of the saturated vapor pressures PSi of pure

propyl acetate and pure ethanoic acid obtained from various experiments
(symbols) and calculations (lines) at dierent temperatures T. Solid
lines, data calculated using the Antoine equation (Table 1); dotted lines,
data calculated using the MPR EoS. For propyl acetate: , this work; g,
ref 27; 3, ref 28; 0, ref 26. For ethanoic acid: , this work; 4, ref 29; O,
ref 12; ], ref 26.

RESULTS AND DISCUSSION

Vapor Pressures of Pure Compounds. To confirm the
reliability of the experimental apparatus and operation and the
purities of the liquid samples, the vapor pressures of pure propyl

Table 1. Critical Pressures (Pc) and Critical Temperatures (Tc),26 Antoine Constants (Ai, Bi, Ci)a,b, Parameters for the MPR EoS
(mi, ni),b and Temperature (T) Range of Application

compound

Pc/MPa

Tc/K

Ai

Bi

Ci

mi

ni

range of T/K

propyl acetate
ethanoic acid

3.360
5.786

549.73
591.95

16.6567
15.3519

4690.84
3700.35

14.864
-45.353

1.00610
1.11336

0.054972
0.047769

323.15 to 359.15
298.15 to 353.15

Antoine equation: ln(PSi /kPa) = Ai - Bi/[(T/K) Ci]. b Obtained from ts to the experimental saturated vapor pressures PSi .
1325

dx.doi.org/10.1021/je101131x |J. Chem. Eng. Data 2011, 56, 13231329

Journal of Chemical & Engineering Data

ARTICLE

Table 2. Total Feeding Amount (nT), Total Feeding Mole Fraction of Propyl Acetate (z1), Vapor-Phase Volume (VG),
Equilibrium Temperature (T) and Pressure (Pexptl), and VLE Data Calculated Using Activity Coecient Models for Propyl
Acetate (1) Ethanoic Acid (2) Binary Mixtures over the T Range (323.15 to 353.15) K
original experimental data
VG

nT
mol

z1

NRTL
Pexptl

cm3

kPa

Wilson
Pcalcd

x1

y1

kPa

UNIQUAC
Pcalcd

x1

y1

kPa

Pcalcd
x1

y1

kPa

T/K = 323.15
0.2782

0.0000

27.37

7.567

0.0000

0.0000

7.626

0.0000

0.0000

7.626

0.0000

0.0000

7.626

0.3132

0.0549

23.73

8.750

0.0549

0.1560

8.566

0.0549

0.1570

8.582

0.0549

0.1565

8.575

0.2525

0.1023

28.07

9.013

0.1022

0.2505

9.224

0.1022

0.2499

9.230

0.1022

0.2499

9.225

0.2389

0.1914

27.79

10.255

0.1913

0.3789

10.220

0.1913

0.3767

10.207

0.1913

0.3772

10.208

0.1948
0.1892

0.3050
0.4091

29.56
28.19

11.316
12.025

0.3049
0.4090

0.4988
0.5897

11.238
12.047

0.3049
0.4090

0.4974
0.5897

11.220
12.038

0.3049
0.4090

0.4978
0.5899

11.223
12.040

0.1697

0.4952

28.99

12.730

0.4951

0.6580

12.671

0.4951

0.6590

12.673

0.4951

0.6590

12.674

0.1586

0.6095

29.17

13.288

0.6094

0.7431

13.467

0.6094

0.7445

13.481

0.6094

0.7443

13.480

0.1577

0.6967

28.05

14.245

0.6966

0.8049

14.061

0.6966

0.8059

14.080

0.6966

0.8058

14.079

0.1476

0.8077

28.24

14.669

0.8076

0.8800

14.811

0.8076

0.8802

14.829

0.8076

0.8802

14.828

0.1796

0.8463

23.59

15.270

0.8463

0.9052

15.070

0.8463

0.9051

15.087

0.8463

0.9051

15.086

0.1622

0.9337

25.17

15.322

0.9337

0.9602

15.658

0.9337

0.9599

15.666

0.9337

0.9599

15.666

0.1527

1.0000

25.73

16.092

1.0000

1.0000

16.103

1.0000

1.0000

16.103

1.0000

1.0000

16.103

0.2782

0.0000

27.40

12.045

0.0000

0.0000

12.114

0.0000

0.0000

12.114

0.0000

0.0000

12.114

0.3132

0.0549

23.92

13.515

0.0549

0.1337

13.239

0.0549

0.1339

13.243

0.0549

0.1338

13.242

0.2525

0.1023

26.72

13.891

0.1022

0.2253

14.112

0.1022

0.2254

14.115

0.1022

0.2254

14.115

0.2389

0.1914

26.23

15.502

0.1913

0.3600

15.557

0.1913

0.3599

15.558

0.1913

0.3600

15.560

0.1948

0.3050

29.63

17.196

0.3048

0.4896

17.129

0.3048

0.4894

17.127

0.3048

0.4895

17.130

0.1892

0.4091

28.10

18.485

0.4089

0.5853

18.386

0.4089

0.5852

18.385

0.4089

0.5853

18.387

0.1697
0.1586

0.4952
0.6095

29.09
29.17

19.347
20.281

0.4950
0.6093

0.6548
0.7388

19.333
20.498

0.4950
0.6093

0.6548
0.7389

19.333
20.498

0.4950
0.6093

0.6547
0.7388

19.334
20.498

0.1577

0.6967

28.04

21.564

0.6966

0.7991

21.333

0.6966

0.7992

21.334

0.6966

0.7991

21.334

0.1476

0.8077

28.26

22.207

0.8076

0.8734

22.345

0.8076

0.8735

22.347

0.8076

0.8734

22.345

0.1796

0.8463

23.67

22.803

0.8462

0.8989

22.685

0.8462

0.8989

22.687

0.8462

0.8989

22.686

0.1622

0.9337

25.14

23.155

0.9337

0.9564

23.438

0.9337

0.9563

23.439

0.9337

0.9564

23.438

0.1527

1.0000

25.66

24.006

1.0000

1.0000

23.993

1.0000

1.0000

23.993

1.0000

1.0000

23.993

T/K = 333.15

T/K = 343.15
0.2782

0.0000

27.39

18.662

0.0000

0.0000

18.655

0.0000

0.0000

18.655

0.0000

0.0000

18.655

0.3132
0.2525

0.0549
0.1023

23.97
26.68

20.354
20.874

0.0549
0.1022

0.1181
0.2054

20.011
21.131

0.0549
0.1022

0.1183
0.2055

20.016
21.138

0.0549
0.1022

0.1170
0.2047

19.980
21.097

0.2389

0.1914

26.29

22.875

0.1913

0.3420

23.113

0.1913

0.3419

23.117

0.1913

0.3428

23.100

0.1948

0.3050

29.52

25.502

0.3048

0.4790

25.411

0.3048

0.4787

25.408

0.3048

0.4807

25.430

0.1892

0.4091

28.15

27.712

0.4089

0.5804

27.306

0.4089

0.5800

27.299

0.4089

0.5816

27.336

0.1697

0.4952

29.12

28.694

0.4949

0.6522

28.731

0.4949

0.6519

28.721

0.4949

0.6527

28.755

0.1586

0.6095

29.15

30.202

0.6092

0.7366

30.441

0.6092

0.7365

30.432

0.6092

0.7362

30.449

0.1577

0.6967

28.00

31.864

0.6965

0.7957

31.620

0.6965

0.7957

31.614

0.6965

0.7950

31.615

0.1476
0.1796

0.8077
0.8463

28.22
23.69

32.815
33.265

0.8076
0.8462

0.8682
0.8934

32.978
33.415

0.8076
0.8462

0.8683
0.8935

32.975
33.413

0.8076
0.8462

0.8677
0.8932

32.962
33.398

0.1622

0.9337

25.06

34.160

0.9337

0.9523

34.334

0.9337

0.9523

34.333

0.9337

0.9524

34.322

0.1527

1.0000

25.73

34.988

1.0000

1.0000

34.962

1.0000

1.0000

34.962

1.0000

1.0000

34.962

T/K = 353.15
0.2782

0.0000

27.41

28.205

0.0000

0.0000

27.933

0.0000

0.0000

27.933

0.0000

0.0000

27.933

0.3132

0.0549

24.02

29.951

0.0549

0.1073

29.590

0.0549

0.1084

29.634

0.0549

0.1050

29.497

0.2525

0.1023

26.75

30.653

0.1022

0.1903

31.010

0.1022

0.1915

31.073

0.1022

0.1886

30.898

0.2389

0.1914

26.77

33.020

0.1912

0.3260

33.635

0.1912

0.3261

33.695

0.1912

0.3274

33.569

1326

dx.doi.org/10.1021/je101131x |J. Chem. Eng. Data 2011, 56, 13231329

Journal of Chemical & Engineering Data

ARTICLE

Table 2. Continued
original experimental data
VG

nT
mol

NRTL

z1

Wilson

Pexptl

cm3

kPa

Pcalcd
x1

y1

UNIQUAC
Pcalcd

kPa

x1

y1

Pcalcd

kPa

x1

y1

kPa

0.1948

0.3050

29.59

36.985

0.3047

0.4677

36.841

0.3047

0.4663

36.858

0.3047

0.4713

36.875

0.1892
0.1697

0.4091
0.4952

28.02
29.23

40.602
41.618

0.4088
0.4948

0.5740
0.6488

39.584
41.674

0.4088
0.4948

0.5720
0.6471

39.556
41.622

0.4088
0.4948

0.5771
0.6505

39.670
41.759

0.1586

0.6095

29.13

43.973

0.6091

0.7349

44.166

0.6091

0.7340

44.108

0.6091

0.7344

44.205

0.1577

0.6967

27.89

46.054

0.6964

0.7935

45.835

0.6964

0.7935

45.791

0.6964

0.7921

45.829

0.1476

0.8077

28.34

47.430

0.8075

0.8642

47.660

0.8075

0.8648

47.645

0.8075

0.8630

47.612

0.1796

0.8463

23.59

47.501

0.8462

0.8889

48.214

0.8462

0.8896

48.209

0.8462

0.8882

48.161

0.1622

0.9337

25.08

49.298

0.9337

0.9482

49.294

0.9337

0.9484

49.302

0.9337

0.9485

49.251

0.1527

1.0000

25.57

49.918

1.0000

1.0000

49.913

1.0000

1.0000

49.913

1.0000

1.0000

49.913

Table 3. Binary Interaction Parameters for Activity Coecient Models and Relative Deviation Percent for Pressure ()a
NRTL
T

(g12-g22)/R

K
323.15
333.15

(g21-g11)/R

-228.181
-84.770

450.343
207.802

Wilson

%
1.20
0.86

(12-11)/R

(21-22)/R

132.233
-79.293

83.552
198.777

UNIQUAC

%
1.21
0.87

(u12-u22)/R
K
-14.515
113.368

(u21-u11)/R
K
81.194
-44.972

%
1.21
0.86

343.15

243.228

-116.766

0.81

-299.481

418.389

0.81

238.204

-132.182

0.79

353.15

466.643

-257.926

0.92

-411.401

602.184

0.93

330.924

-183.490

0.89

/% = (100/Np)

PNp

calcd
i = 1|Pi

- Pexptl
|/Pexptl
, where Np is the number of data points.
i
i

acetate and ethanoic acid were first determined. Figure 1 shows

our results for the experimental vapor pressures of pure propyl
acetate and ethanoic acid over the T = (323.15 to 359.15) K and
(298.15 to 353.15) K ranges, respectively. Experimental vapor
pressures from the literature12,26-29 and ones calculated using
the Antoine equation and the MPR EoS are also provided in the
figure. It can be seen that our measurements are in excellent
agreement with the literature data, suggesting that the apparatus
was working well and that the pure liquid samples were sufficiently pure to be further used for the measurements on mixtures.
VLE Data for Binary Mixtures. As mentioned above, the
original experimental values are nT, zi, T, Pexptl, and VG. From
these original data, one can obtain the complete phase diagram
with the help of simple thermodynamic models and mass-balance
equations.
The mass-balance equations lead to
xi

zi - f 3 yi
1-f

14

where yi is the vapor-phase mole fraction and f is the vapor

fraction, which is given by
VG
f G
15
Vm 3 nT
in which VmG is the molar volume of the vapor phase estimated
using the MPR EoS.10,11 The required values of the critical
pressure Pc and temperature Tc for the MPR EoS were taken

from the literature26 and are listed in Table 1. Moreover, the

values of the parameters m and n in the MPR EoS and the
constants Ai, Bi, and Ci in the Antoine equation obtained by
tting the experimental vapor pressures of pure substances are
also reported in Table 1.
The VLE condition is given by
yi 3 P 3 jVi xi 3 i 3 PiS 3 jSi 3 expViL 3 P - PiS =R 3 T

16

where PSi is the vapor pressure of component i, jVi and jSi are the
fugacity coecients of component i in the vapor mixture and the
pure state, respectively (calculated using the MPR EoS),10,11 and
i is the activity coecient of component i in the liquid phase
(calculated using the activity coecient models). It should be
noted that the parameter Rij in the NRTL model was set to 0.3
here. The term exp[VLi 3 (P - PSi )/(R 3 T)] is the Poynting
correction factor, which was set to 1 in the current work because
the experimental pressure was low. Equations 14 to 16 were
solved iteratively by self-substitution. The binary parameters of
the activity coecient models were optimized by using the
Nelder-Mead simplex method30 and the following objective
function F:


Np  calcd
exptl 
X
Pi - Pi 
17
F


exptl


Pi
i1
where Pcalcd and Pexptl denote the calculated and experimental
equilibrium pressures and Np is the number of data points.
1327

dx.doi.org/10.1021/je101131x |J. Chem. Eng. Data 2011, 56, 13231329

Journal of Chemical & Engineering Data

ARTICLE

temperatures are depicted in Figure 2. It should be noted that

only the results calculated using the NRTL model are given in
this gure, since the Wilson and UNIQUAC models produced
the same results as NRTL and the calculated pressures obtained
from NRTL are in good agreement with the experimental
equilibrium pressures. Furthermore, the binary parameters for
all of the models can be linearly correlated as functions of temperature. The linear expressions for each model are given below:
For the NRTL model:
g12 -g22 =R 24:1247 3 T - 8058:54
g21 -g11 =R -24:4937 3 T 8353:42

18a
18b

For the Wilson model:

12 -11 =R -18:5109 3 T 6094:98

19a

21 -22 =R 17:7551 3 T - 5678:15

19b

For the UNIQUAC model:

Figure 2. Comparison of binary VLE data for the propyl acetate (1)
ethanoic acid (2) system at various temperatures obtained from experiment (symbols) and theoretical calculations and predictions (lines). (a)
353.15 and 343.15 K; (b) 333.15 and 323.15 K. The generated
coexistence compositions were determined using MPR NRTL model.
], 353.15 K; 4, 343.15 K; 0, 333.15 K; O, 323.15 K; solid lines, data
calculated using NRTL; dashed lines, data predicted by the new version
of COSMO-SAC;14 dotted lines, data predicted by the original
COSMO-SAC.21

The experimental temperature T, the total feeding amount nT,

the total mole fraction of liquid feeding z1 for propyl acetate, the
vapor-phase volume VG, the experimental pressure Pexptl, and the
calculated liquid- and vapor-phase mole fractions x1 and y1 for
propyl acetate, and the pressure Pcalcd are reported in Table 2.
The binary parameters of the NRTL, Wilson, and UNIQUAC
models and the corresponding average deviations can be found in
Table 3. It can be seen from Table 2 that the composition of the
liquid phase at equilibrium, x1, was very similar to the total
composition of the liquid feeding, z1, which indicates that only a
small amount of the feeding liquid sample was vaporized (i.e., the
vapor fraction was small). This is due to the fact that the
vapor-phase volume was very small (about 27 cm3) after injection of the liquid sample (about 18 cm3) into the equilibrium still
(45 cm3). Furthermore, the three dierent activity coecient
models gave about the same coexistence compositions and
equilibrium pressures, and the average deviations were almost
same for the three models. The pressure-composition diagrams
of the propyl acetate ethanoic acid mixture at four dierent

u12 -u22 =R 11:6115 3 T - 3759:44

20a

u21 -u11 =R -8:81260 3 T 2910:13

20b

As a result, the isobaric or isothermal VLE data of propyl acetate ethanoic acid mixtures over the T range (323.15
to 353.15) K can be smoothly predicted using these
equations.
Predictions Using COSMO-SAC Models. In COSMO-like
activity coefficient models, the only required input that characterizes a given compound is the so-called COSMO file, which is
the output file from a COSMO ab initio quantum calculation. In
this work, the COSMO files for propyl acetate and ethanoic acid
were those from the VT 2005 database developed by Mullins et
al.22 The predicted pressure-composition diagrams are shown
in Figure 2. It can be clearly seen that the predictions of the new
version of COSMO-SAC14 are more accurate than those of the
original COSMO-SAC model21 for the investigated mixture.
This seems to indicate that the use of separate different profiles
to distinguish hydrogen-bonding and non-hydrogen-bonding
surface segments leads to a better representation of the propyl
acetate and ethanoic acid associating system.

CONCLUSIONS
VLE data for propyl acetate ethanoic acid binary mixtures
over the T range (323.15 to 353.15) K have been determined
using the static total pressure method and thermodynamic
correlations. The NRTL, Wilson, and UNIQUAC activity coecient models have been used to represent the nonideality of the
liquid phase, and the MPR EoS has been used to describe the
properties of the vapor phase. The VLE data are well-correlated
by the three activity coecient models, and the overall average
relative deviations from the equilibrium pressures are (0.95, 0.96,
and 0.94) % for the NRTL, Wilson, and UNIQUAC models,
respectively. The predictive COSMO-SAC approach has been
used to represent the binary VLE data of the propyl acetate
ethanoic acid system, and good agreement with the experimental
data was obtained using the new version of COSMO-SAC.
1328

dx.doi.org/10.1021/je101131x |J. Chem. Eng. Data 2011, 56, 13231329

Journal of Chemical & Engineering Data

AUTHOR INFORMATION
Corresponding Author

*Tel.: 86 021 64252767. E-mail: cjpeng@ecust.edu.cn.

Funding Sources

The authors from East China University of Science and Technology appreciate the nancial support of the National Natural
Science Foundation of China (20876041), the National Basic
Research Program of China (2009CB219902), the Program for
Changjiang Scholars and Innovative Research Team in University of China (Grant IRT0721), and the 111 Project of China
(Grant B08021). J.L. and P.P. thank the French National Agency
for Research (ANR-09-CP2D-10-03) for nancial support.

ACKNOWLEDGMENT
We are grateful to Lin and co-workers for their help in
reproducing the results of the 2010 version of the COSMOSAC model.
REFERENCES
(1) Kirk, R. E. Encyclopedia of Chemical Technology, 3rd ed.; WileyInterscience: New York, 1978.
(2) Othmer, D. F. Process for Dehydration of Acetic Acid and Other
Lower Fatty Acids. U.S. Patent 2,076,184, April 6, 1937.
(3) Othmer, D. F. Composition of Vapors from Boiling Binary
Solutions. Ind. Eng. Chem. 1943, 35, 614620.
(4) Fu, H.; Chen, G. H.; Han, S. J. The vapor liquid equilibrium of
systems containing associating compounds II: Binary VLE of acetic acid
with water, 2-butanone, 2-pentanone and ethyl acetate. Chem. Eng.
(China) 1986, 5661.
(5) Xiao, X. J.; Wang, L. J.; Ding, G. H.; Li, X. Liquid-liquid
equilibria for the tertiary system water acetic acid propyl acetate.
J. Chem. Eng. Data 2006, 51, 582583.
(6) Toikka, A.; Toikka, M. Phase transitions and azeotropic properties of acetic acid-n-propanol-water-n-propyl acetate system. Fluid
Phase Equilib. 2006, 250, 9398.
(7) Renon, H.; Prausnitz, J. M. Local compositions in thermodynamic excess functions for liquid mixtures. AIChE J. 1968, 14, 135144.
(8) Wilson, G. M. Vapor-Liquid Equilibrium. XI. A New Expression for the Excess Free Energy of Mixing. J. Am. Chem. Soc. 1964, 86,
127130.
(9) Abrams, D. S.; Prausnitz, J. M. Statistical thermodynamics of
liquid mixtures: A new expression for the excess Gibbs energy of partly
or completely miscible systems. AIChE J. 1975, 21, 116128.
(10) Liu, H. L.; Feng, G. X.; Hu, Y. A modied PR equation of state I:
Application to pure uid. Nat. Gas Chem. Ind. (China) 1987, No. 1, 4751.
(11) Liu, H. L.; Feng, G. X.; Hu, Y. A modied PR Equation of state II:
Allication to mixture. Nat. Gas Chem. Ind. (China) 1987, No. 2, 4854.
(12) Hu, Y.; Wang, K.; Lu, R. D. Study on the Isothermal Vapor
Liquid Equilibrium of Acetic Acid-Water Binary over the Temperature
Range 25-75C. J. Chem. Ind. Eng. (China) 1980, 341351.
(13) Sun, L.; Fu, J. Y.; Li, W.; Zhan, X. L.; Xiao, W. D. Binary
vapor-liquid equilibrium of methyl glycolate and ethylene glycol. Fluid
Phase Equilib. 2006, 250, 3336.
(14) Hsieh, C. M.; Sandler, S. I.; Lin, S. T. Improvements of
COSMO-SAC for vapor-liquid and liquid-liquid equilibrium predictions. Fluid Phase Equilib. 2010, 297, 9097.
(15) Peng, D. Y.; Robinson, D. B. A New Two-Constant Equation of
State. Ind. Eng. Chem. Fundam. 1976, 15, 5964.
(16) Klamt, A. Conductor-like Screening Model for Real Solvents: A
New Approach to the Quantitative Calculation of Solvation Phenomena.
J. Phys. Chem. 1995, 99, 22242235.
(17) Klamt, A.; Jonas, V.; Burger, T.; Lohrenz, J. C. W. Renement and
Parametrization of COSMO-RS. J. Phys. Chem. A 1998, 102, 50745085.

ARTICLE

(18) Klamt, A.; Eckert, F. COSMO-RS: A novel and ecient

method for the a priori prediction of thermophysical data of liquids.
Fluid Phase Equilib. 2000, 172, 4372.
(19) Klamt, A.; Krooshof, G. J. P.; Taylor, R. COSMOSPACE:
Alternative to conventional activity-coecient models. AIChE J. 2002,
48, 23322349.
(20) Klamt, A. COSMO-RS: From Quantum Chemistry to Fluid Phase
Thermodynamics and Drug Design; Elsevier: Amsterdam, 2005.
(21) Lin, S. T.; Sandler, S. I. A priori phase equilibrium prediction
from a segment contribution solvation model. Ind. Eng. Chem. Res. 2002,
41, 899913.
(22) Mullins, E.; Oldland, R.; Liu, Y. A.; Wang, S.; Sandler, S. I.;
Chen, C. C.; Zwolak, M.; Seavey, K. C. Sigma-prole database for using
COSMO-based thermodynamic methods. Ind. Eng. Chem. Res. 2006, 45,
43894415.
(23) Wang, S.; Sandler, S. I.; Chen, C. C. Renement of COSMOSAC and the applications. Ind. Eng. Chem. Res. 2007, 46, 7275
7288.
(24) Lin, S. T.; Chang, J.; Wang, S.; Goddard, W. A.; Sandler, S. I.
Prediction of vapor pressures and enthalpies of vaporization using a
COSMO solvation model. J. Phys. Chem. A 2004, 108, 74297439.
(25) Wang, S.; Lin, S. T.; Chang, J.; Goddard, W. A.; Sandler, S. I.
Application of the COSMO-SAC-BP solvation model to predictions
of normal boiling temperatures for environmentally signicant substances. Ind. Eng. Chem. Res. 2006, 45, 54265434.
(26) Perrys Chemical Engineers Handbook, 8th ed.; Green, D. W.,
Perry, R. H., Eds.; McGraw-Hill: New York, 2008.
(27) Ambrose, D.; Ellender, J. H.; Gundry, H. A.; Lee, D. A.;
Townsend, R. Thermodynamic properties of organic oxygen compounds LI. The vapour pressures of some esters and fatty acids. J. Chem.
Thermodyn. 1981, 13, 795802.
(28) Polak, J.; Mertl, I. Saturated Vapor Pressure of Methyl Acetate,
Ethyl Acetate, n-Propyl Acetate, Methyl Propionate, and Ethyl Propionate. Collect. Czech. Chem. Commun. 1965, 30, 35263528.
(29) McDonald, R. A.; Shrader, S. A.; Stull, D. R. Vapor Pressures
and Freezing Points of 30 Organics. J. Chem. Eng. Data 1959, 4, 311
313.
(30) Nelder, J. A.; Mead, R. A simplex method for function minimization. Comput. J. 1965, 7, 308313.

1329

dx.doi.org/10.1021/je101131x |J. Chem. Eng. Data 2011, 56, 13231329